Deciphering Lobular Breast Cancer: A Distinct Entity — Clinical Pitfalls, Biomarker Limitations, and Individualised Treatment Decisions
Blog Navigating ER+HER2− Early Breast Cancer
Meritxell Bellet, MD, PhD • Mariangela Gaudio, MD
Vall d’Hebron University Hospital / VHIO / SOLTI, Barcelona | Eurobio Scientific Webinar, 23 April 2026
Abstract
Invasive lobular carcinoma (ILC) accounts for 8–15% of all breast cancers and is now recognised as a distinct biological entity. At an international webinar on 23 April 2026, Dr Meritxell Bellet (VHIO/SOLTI, Barcelona) and Dr Mariangela Gaudio guided a breast cancer specialist audience through the current evidence landscape using a real-world patient case as a running thread. This article explores the molecular biology of ILC, discusses the limitations of the Ki67 threshold and the ADAPT algorithm in this histological subtype, and compares the commercially available multigene signatures with respect to their validity in lobular carcinoma. The case of a 77-year-old patient with cT3N0 ILC illustrates how two genomic platforms (Oncotype DX® and EndoPredict®/EPclin) can yield divergent risk assessments, and how clinical, pathological, and patient-specific factors together support an individualised adjuvant decision.
Key message: Clinicians should consistently treat ILC as a separate entity, apply ILC-specific Ki67 thresholds, and select genomic platforms based on validated performance in the lobular subtype.
Introduction: ILC as a Distinct Biological Entity
The medical and scientific understanding of invasive lobular carcinoma (ILC) has experienced a paradigm shift as the oncology community increasingly moves away from treating it as a subset of hormone receptor-positive breast cancer and toward recognizing it as a distinct biological entity with unique clinical challenges.
ILC accounting for 8–15% of all HR+/HER2- breast cancers, is the second most common histological subtype — and yet it continues to be treated in many clinical settings as a mere variant of Invasive Carcinoma of No Special Type (NST/IDC). Using a clinical case during the latest webinar Dr Meritxell Bellet (VHIO/SOLTI, Barcelona) and Dr Mariangela Gaudio, clearly demonstrated, that ILC differs from IDC in fundamental ways across pathology, molecular biology, staging behaviour, and therapeutic response.
History and Diagnosis
Dr Gaudio presented the case of a 77-year-old woman (ECOG 0) with a background of dyslipidaemia, cutaneous lupus (anti-Ro52/Ro60+), osteoporosis (T-score −2.5), and a positive family history (sister: lymphoma and breast cancer at age 73). No prior hormone therapy; BRCA status unknown.
In November 2022, the patient self-detected a lump in her left breast. Strikingly, initial mammography and ultrasound screening yielded only a BI-RADS 2 result — a classic illustration of the well-known imaging pitfall of ILC in dense breast tissue. Only after nodule growth in May 2023 (BI-RADS 4A) did a core biopsy establish the diagnosis: Invasive Lobular Carcinoma Grade 1, ER 100%, PR 85%, HER2 2+/ISH−, Ki67 13%.
Pre-surgical imaging (MRI, CT thorax/abdomen, bone scan) revealed a 56 mm non-mass enhancement lesion in the internal upper quadrant with signs of muscular infiltration (BI-RADS 6) — no evidence of distant metastases. Staging: cT3 cN0 cM0.
Treatment decision: Neoadjuvant endocrine therapy (NET) with letrozole — a sound initial approach for a classically “Luminal A-like” ILC, as the subsequent course would confirm.
Neoadjuvant Endocrine Therapy and Response
After seven months of letrozole, the MRI in January 2024 demonstrated a complete radiological response — no residual tumour was identifiable from the initial 56 mm composite lesion. The formal classification remained clinical partial response.
Surgery and Histopathology (February 2024)
The patient underwent a TOPL (crescent mastopexy) including pectoral muscle fibres and sentinel lymph node biopsy. Pathological work-up revealed:
- Residual ILC Grade 2, largest focus 12 mm (ypT1c)
- Multiple foci of pleomorphic LCIS (apocrine pattern, nuclear grade 3) comprising 80% of the residuum
- Sentinel nodes 0/3 negative (ypN0), no LVI
- ER post-NET 90%, Ki67 post-NET 1%, PR 0%, HER2 0 (IHC)
- RCB 1.02 — Class I | PEPI Score 0 (favourable response)
A PEPI score (Post-neoadjuvant Endocrine Prognostic Index) of 0 points reflects excellent endocrine sensitivity and is associated with a very favourable prognosis. The patient’s biological response to NET was compelling — and would prove pivotal to the subsequent adjuvant decision.
Pathological Differentiation from NST
The defining biological hallmark of invasive lobular carcinoma is the loss of cell-cell adhesion, a characteristic driven almost universally by the inactivation of the E-cadherin protein. E-cadherin, encoded by the CDH1 gene located on chromosome 16q22, acts as the primary “molecular glue” that allows epithelial cells to form cohesive structures. The mechanical consequence of this molecular loss is the characteristic discohesive growth pattern. ILC cells do not form palpable lumps or distinct masses; instead, they infiltrate the surrounding breast tissue in thin, single-file lines or small chains, often referred to as “Indian filing”. This growth pattern allows the cancer to spread diffusely through the breast stroma without significantly disrupting the normal architecture, which contributes to its “sneaky” nature and the high frequency of under-diagnosis by traditional screening methods.
Biologically, ILC is almost universally HR-positive and HER2-negative, displays lower histological grades and lower Ki67 values, tends towards multifocal and multicentric disease, and exhibits a characteristic pattern of distant metastasis (peritoneum, pleura, ovaries, bone marrow, meninges).
Molecular Signature: Mutations and DNA Profile
ILC carries an unmistakable molecular fingerprint: CDH1 mutations (~65–68%) are near-universal; PIK3CA mutations (~47–55%) are more frequent than in IDC; ERBB2 mutations (~5–8%) are associated with inferior distant relapse-free survival and are enriched in CDH1-mutated tumours.
At the chromosomal level, 1q gain (~74–84%), 16q loss (~63–84%), and 22q loss (~50%) are ILC-characteristic. Crucially, the overall genomic complexity of ILC is lower than that of IDC. Where IDC frequently shows structural rearrangements and HRD-positive profiles, elevated TMB in ILC is primarily APOBEC-driven — a fundamentally different pattern with potential therapeutic implications.
Prognostic Factors in Early ILC
A distinctive feature of the ILC survival curve is the “crossover” effect observed in long-term outcomes.
Classical adverse prognostic features include pT > 5 cm, pN+, non-Luminal A-like subtypes, non-classic histologies, and age > 70 years. Since ILC is notably prone to late recurrences, this is an aspect that should be systematically considered when planning the duration of extended endocrine therapy.
One of the most clinically impactful points of the webinar was the detailed discussion of Ki67 thresholds. The guideline-endorsed cut-off of 14–20% to distinguish “Luminal A” from “Luminal B” tumours is not valid in ILC — it systematically underestimates the risk of recurrence.
A study of more than 1,000 patients (Carbognin et al., The Breast, 2017) demonstrated that the optimal discriminatory Ki67 threshold for ILC is as low as 4% — with significant DFS and OS separation above this cut-off in both training and validation cohorts. For IDC, the established 14% threshold remained valid.
New data from the VHIO/SOLTI group (ESMO Breast 2026, Abstract P107): in a cohort of 79 pure ILC patients, Ki67 correlated significantly with the EndoPredict Molecular Score (Spearman rho = 0.57; p < 0.001). The optimal stratification threshold was 11% — well below current guideline recommendations.
The practical implication is clear: a patient with ILC and Ki67 = 13% — as in our case — is classified as “Luminal A-like” by current guideline criteria. Biologically, however, this value in ILC may correspond to a substantially higher recurrence risk than the same Ki67 value in IDC. The Luminal A concept is not directly transferable to ILC.
The ADAPT Algorithm and Its Limitations in ILC
The WSG-ADAPT concept (Nitz et al., JCO, 2022) has established an elegant adaptive treatment algorithm for HR+/HER2− early breast cancer: a 3-week short-course ET trial is combined with Recurrence Score (RS) measurement from the initial biopsy.
Patients with initial RS 12–25 and Ki67 drop to ≤10% on second biopsy achieved outcomes comparable to the control arm.
When applied to ILC, however, this concept has several critical limitations:
- Low baseline Ki67: Most ILC tumours already have a Ki67 below 10% at baseline. A further drop below this threshold therefore does not necessarily reflect a true treatment response — it may simply reflect the intrinsically low proliferative nature of the tumour.
- Multifocality/multicentricity: ILC frequently presents with multifocal or multicentric growth, with potentially different phenotypic and genomic properties across lesions.
- Only Oncotype DX studied: The ADAPT algorithm relies on the Recurrence Score — a platform with significant limitations in ILC (see below). No published ILC-specific subgroup analyses from the ADAPT programme are available.
Conclusion: The ADAPT algorithm cannot be directly transferred to ILC. Given the already low baseline Ki67, the systematic RS underestimation by Oncotype DX, and the multifocality issue, risk stratification in ILC requires a histology-specific approach (Christgen et al.; Cancer, 2020).
Consistent signals come from SOFT/TEXT and the TEAM trial. Aromatase inhibitors are the clear first choice for adjuvant endocrine therapy in ILC. Given the late-recurrence tendency of ILC, extended endocrine therapy beyond five years should be regularly considered. Data on new oral SERDs (elacestrant, giredestrant) are awaited with interest.
The systemic treatment of invasive lobular carcinoma is fundamentally different from that of ductal carcinoma, primarily because ILC is almost invariably hormone receptor-positive and thought to be less responsive to conventional chemotherapy. Endocrine therapy remains the backbone of management for nearly all patients with early-stage ILC.
An exploratory post-hoc subgroup analysis of the PENELOPE-B trial (n = 110 ILC) showed a trend towards improved iDFS and OS with palbociclib after nearly seven years of follow-up — statistically significant for iDFS. The PALLAS study likewise yielded a more favourable HR for the ILC subgroup than for the IDC cohort, albeit with a weak interaction.
ILC-specific subgroup analyses for ribociclib (NATALEE), abemaciclib (monarchE), and ribociclib in the RIBOLARIS trial are still awaited.
In the presented case, ribociclib under the NATALEE protocol (cT3N0, EPclin high-risk) was considered as an option — at the time of the decision, pending approval status.
For decades, the standard narrative has been that invasive lobular carcinoma is “chemoresistant” because it achieves significantly lower rates of pathologic complete response (pCR) in the neoadjuvant setting compared to IDC. However, the current clinical perspective, spearheaded by leading researchers like Dr. Otto Metzger, argues that this oversimplifies the biology of the disease.
Multiple meta-analyses and systematic reviews — predominantly based on retrospective data — show no OS benefit from adjuvant chemotherapy in ILC. A pooled analysis of over 38,000 ILC patients (Trapani et al., 2021) found no survival advantage; a second large meta-analysis (Davey et al., 2022) confirmed the absence of long-term DFS and OS benefit.
Data from the National Cancer Database even showed that neoadjuvant chemotherapy in ILC — compared with adjuvant chemotherapy — was associated with worse overall survival, without delivering the expected advantages (lower mastectomy rates, less axillary surgery).
A retrospective study (de Nonneville et al., 2019) did identify a DFS and OS benefit from chemotherapy in clearly defined high-risk ILC patients. Anthracycline-containing regimens appear to provide added value over taxane monotherapy only in the pN2/3 subgroup.
Clinical conclusion: Chemotherapy in ILC should be reserved for clearly defined high-risk patients (pN2/3 or pN0/1 with high genomic risk by EPclin or MammaPrint). There is no evidence base for its routine use in genomically low-risk ILC.
The choice of genomic platform is not arbitrary in ILC. Dr Bellet emphasised that Oncotype DX has significant limitations in this histology, and recommended a clear hierarchy:
EPclin > MammaPrint ≥ Prosigna > Oncotype DX
Overview of commercially available signatures in the ILC context:
| Platform | Genes (n) | % ILC High-Risk | ILC Validation | Key ILC-Specific Limitation / Finding |
| EndoPredict® (EPclin) | 12 + T + N | ~37% | Yes (470 pts, 3 RCTs) | Best-validated platform for ILC; HR 3.32; integrates tumour size + nodal status; significant predictor of chemotherapy benefit |
| MammaPrint (70-gene) | 70 | ~11-16% | Partial (MINDACT) | Better than Oncotype DX; trend towards CT benefit in genomically high-risk ILC; no ILC-specific validation study |
| Prosigna (PAM50) | 50 | ~31% (Low ROR 41%) | Partial (ABCSG-8) | Underestimates ILC risk: Low-ROR ILC has significantly higher 10-yr DR than Low-ROR IDC (7.7% vs 3.5%) |
| Oncotype DX® (RS) | 21 | 2–3.6% | No | Strongly driven by oestrogen signalling (ρ = −0.79); clusters ILC into Low/Intermediate RS; not prognostic in ILC (WSG PlanB multivariate analysis) |
Oncotype DX® (21-Gene RS)
The RS algorithm is strongly oriented towards oestrogen signalling genes (correlation RS–ER activity: ρ = −0.79). Because ILC tumours are almost universally strongly ER-positive, they systematically cluster into the Low/Intermediate RS range — only 2–4% of ILC tumours achieve RS ≥26, compared with 16–24% of IDC. In the WSG PlanB trial, the RS was not an independent prognostic factor for ILC in multivariate analysis — in contrast to IDC.
EndoPredict® / EPclin
EndoPredict is the best-validated commercial platform for ILC. In a pooled analysis of 470 ILC patients from three phase III trials (Šestak et al., Clin Cancer Res, 2020), EPclin demonstrated an HR of 3.32 in ILC — consistent across N0 and N+ patients and across ET-alone and ET+CT-treated patients. Notably, 37% of ILC patients were classified as EPclin high-risk (vs only ~9.5% with Oncotype DX), and EPclin was a significant predictor of chemotherapy benefit (p-interaction = 0.022).
One methodological pitfall deserves attention: EPclin integrates molecular score + tumour size + nodal status. In the presented case (primary tumour size 56 mm/cT3), the EPclin score was partly driven by the clinical tumour size — a consideration that warrants careful interpretation in the post-NET setting.
MammaPrint (70-Gene) and Prosigna (PAM50)
MammaPrint classifies fewer ILC patients as genomically high-risk than NST (11–16% vs 39%) and shows a trend towards CT benefit in the genomically high-risk group — without an ILC-specific validation study.
Prosigna classifies 41% of ILC as Low-ROR but underestimates their risk: Low-ROR ILC has significantly higher 10-year distant recurrence rates than Low-ROR IDC (7.7% vs 3.5%).
After completion of neo-adjuvant endocrine therapy (NET), surgery (February 2024), and radiotherapy (April 2024), the question of adjuvant systemic therapy arose. The patient explicitly wished for genomic support before any chemotherapy decision. Both available tests were accidently requested from the biopsy material — with divergent results:
| Parameter | Oncotype DX® | EndoPredict® / EPclin |
| Result | RS 7 | MS 5.5 / EPclin 3.6 |
| Risk class | Low Risk | High Risk (cut-off 3.3) |
| 10-yr DR on ET | ~3% (95% CI 2–4%) | ~13% |
| Absolute CT benefit | <1% (RS 0–10, age >50) | ~4% absolute |
| ILC validation | No dedicated study | 470 ILC patients, 3 phase III trials |
| ILC-specific limitation | RS systematically low; not prognostic | EPclin incorporates primary tumour size (cT3) |
The Final Treatment Decision
After multidisciplinary discussion and taking the patient’s preferences into account, the decision was made against chemotherapy and in favour of endocrine therapy alone, based on the following reasoning:
- PEPI Score 0: Excellent pathological response to NET (ypT1c, ypN0, Ki67 post-NET 1%, ER 90%) — confirming endocrine sensitivity
- Oncotype RS 7: Per TAILORx criteria (RS 0–10, age >50), absolute chemotherapy benefit <1%
- EPclin 3.6: Close to cut-off high-risk (cut-off 3.3), partly driven by clinical tumour size of 56 mm (cT3); molecular score alone: 5.5
- Absolute EPclin CT benefit ~4%: Insufficient grounds for chemotherapy in a 77-year-old patient with comorbidities
- Patient profile: Age 77, osteoporosis, cutaneous lupus — overall benefit–risk balance strongly argues against chemotherapy
Dr Bellet concluded: “The patient wanted to be confident that, for a decision of this magnitude, we had genomic support. Both platforms together, combined with the PEPI score and clinical parameters, produced a consistent picture that does not justify escalation to chemotherapy.”
Final decision: Anastrozole 2.5 mg/day · Planned duration 7–8 years · Bone health management (bisphosphonate/RANKL inhibitor) · Ribociclib (NATALEE, cT3N0) under evaluation pending approval status
TBCRC 037 (Window-of-Opportunity Study): This randomised window-of-opportunity study demonstrated in ILC that fulvestrant achieved a greater Ki67 reduction than tamoxifen (p = 0.042), while tamoxifen and anastrozole performed comparably. Proteomic and RNAseq analyses confirmed altered ER activity in ILC — an important foundation for future SERD trials.
ROSALINE Trial: The rationale was the synthetic lethality of E-cadherin-deficient cells to ROS1 kinase inhibition. The phase II study of entrectinib plus letrozole showed limited efficacy (RCB 0–1 in only one patient) alongside substantial toxicity (grade ≥3 events in 25%, dose reductions in 55%).
NCT05919108 (Neratinib Study): A phase II trial evaluating the HER2-TKI neratinib plus endocrine therapy in ERBB2-mutant HR+ ILC — an exploratory approach exploiting the higher prevalence of ERBB2 mutations in ILC compared with IDC.
Take-Home Messages for Clinical Practice
- ILC is a distinct biological entity with specific molecular features (CDH1, 1q gain, 16q loss) and a clinical–radiological behaviour that differs fundamentally from IDC.
- Ki67 thresholds must be reconsidered for ILC: The Luminal A cut-off of 14–20% is not valid for ILC. Current data point to an optimal threshold of approximately 10–11%.
- The choice of genomic test is not arbitrary in ILC: EPclin > MammaPrint ≥ Prosigna > Oncotype DX. Oncotype DX systematically underestimates ILC risk and lacks prognostic validity in this histological subtype.
- The ADAPT algorithm is not directly transferable to ILC: Low baseline Ki67, multicentricity, staging difficulties, and the absence of ILC validation for the sole platform used (RS) substantially limit its applicability.
- Chemotherapy: Reserve for high-risk patients (pN2/3 or pN0/1 with high genomic score by EPclin/MammaPrint). Aromatase inhibitors are the cornerstone of ET; consider extended endocrine therapy.
- Dedicated ILC research is urgently needed: ILC-specific cohorts, novel signatures (LobSig, PSILC), and dedicated trial protocols are essential to adequately serve this underrepresented patient population.
References:
To get even more information on this topic from the exciting discussion of our experts, please watch the recording of the webinar.
https://eurobio.webinargeek.com/deciphering-lobular-breast-cancer
Gene-expression related references:
Buus et al., JCO, 2020
Christgen et al., Cancer, 2020
Gnant et al., Ann Oncol, 2014
Jenkins et al., BCRT, 2022
Kizy et al., BCRT, 2017
Kizy et al., APLM, 2018
Laenkholm et al., CCR, 2020
Metzger et al., JCO, 2015
Metzger et al.,EJC, 2025
Nitz_J et al., Clin Oncol, 2022
Sestak et al., JAMA Oncol, 2018
Sestak et al., BCRT, 2019
Sestak et al., CCR, 2020
Author
Stephanie Ahrendt – Global Medical Scientific Liaison Manager
Stephanie leverages a deep background in molecular biology and clinical pharmacology, alongside extensive experience in oncology, endocrinology, and nuclear medicine. In her current role at Eurobio Scientific, she is dedicated to spearheading innovative oncology therapies and translating scientific discovery into clinical practice.
Key Takeaways
- The article presents a case study of a 77-year-old woman diagnosed with Invasive Lobular Carcinoma (ILC) and discusses her treatment journey.
- Neoadjuvant endocrine therapy using letrozole resulted in a complete radiological response after seven months, leading to a decision against chemotherapy.
- Key findings highlight that ILC has distinct biological and molecular characteristics, requiring different thresholds for Ki67 and genomic test selection compared to invasive ductal carcinoma.
- The ADAPT algorithm is not suitable for ILC due to unique challenges like low baseline Ki67 and multifocality.
- The authors emphasize the urgent need for dedicated research and clinical trials focused on ILC to improve treatment outcomes.
English (UK)